US5317753A - Coordinate rotation digital computer processor (cordic processor) for vector rotations in carry-save architecture - Google Patents

Coordinate rotation digital computer processor (cordic processor) for vector rotations in carry-save architecture Download PDF

Info

Publication number
US5317753A
US5317753A US07/667,289 US66728991A US5317753A US 5317753 A US5317753 A US 5317753A US 66728991 A US66728991 A US 66728991A US 5317753 A US5317753 A US 5317753A
Authority
US
United States
Prior art keywords
vector
carry
angle
output
input
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US07/667,289
Other languages
English (en)
Inventor
Ronald Kuenemund
Tobias Noll
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Infineon Technologies AG
Original Assignee
Siemens AG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens AG filed Critical Siemens AG
Assigned to SIEMENS AKTIENGESELLSCHAFT, A GERMAN CORPORATION reassignment SIEMENS AKTIENGESELLSCHAFT, A GERMAN CORPORATION ASSIGNMENT OF ASSIGNORS INTEREST. Assignors: KUENEMUND, RONALD, NOLL, TOBIAS
Application granted granted Critical
Publication of US5317753A publication Critical patent/US5317753A/en
Assigned to INFINEON TECHNOLOGIES AG reassignment INFINEON TECHNOLOGIES AG ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: SIEMENS AKTIENGESELLSCHAFT
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
    • G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
    • G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
    • G06F7/544—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices for evaluating functions by calculation
    • G06F7/5446—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices for evaluating functions by calculation using crossaddition algorithms, e.g. CORDIC

Definitions

  • the present invention relates to a computer processor, and is more particularly concerned with a coordinate rotation digital computer processor CORDIC processor, for vector rotations for solving problems of real-time processing, constructed with a carry-save architecture.
  • CORDIC coordinate rotation digital computer processor
  • a CORDIC processor is known from the publication of Helmut Halm, et al entitled “CORDIC-Prozessoren fuer die digitale Signaltechnik”, published in the periodical "me", Vol. 3, No. 1, 1989, pp. 22-27.
  • the object of the present invention is to provide a CORDIC processor that is constructed of simple elementary cells, can be easily modified in view of accuracy and word width and mainly represents a good compromise between low chip surface and high data rate.
  • the advantage which may be obtained in practicing the present invention is, in particular, that a significantly-improved relationship of data rate to chip surface requirement occurs in the CORDIC processor constructed in accordance with the present invention in comparison to known CORDIC processors, this resulting from the carry-save architecture, and that the data rate is independent of the overall word width.
  • each angle iteration stage is composed of a plurality of identical angle path base cells; the sign detectors are connected parallel to angle path base cells; the input lines for angle bits of the angle path base cells are connected to lines for the non-inverted and inverted sign output signals of the respectively immediately-preceding angle iteration stage, such that this corresponds to a product formation of the sign output signal of the immediately-preceding angle iteration stage and the binary representation of a respective given, scaled negative angle step, whereby bits not modifiable by the product formation are occupied with fixed logical values; and the structures for realizing a shift operation required for scaling the carry and sum words at the output of an angle iteration stage are comprised such that the output lines for the carry-save bits of an angle path base cell are respectively connected to the outputs of the next more-significant angle path base cell of an immediately-following angle iteration stage.
  • the processor is characterized in that angle path base cells of an angle iteration stage each contain a respective full adder; that a register, respectively, for carry bit and sum bit and clock by clock signals follows the full adder for mutual decoupling of the vector iteration stages.
  • the inputs of the angle path base cell correspond to the inputs of the full adder, whereby one input of the full adder is connected to the input line for an angle bit and two other inputs of the full adder are connected to the output lines for the carry and sum bits of the immediately-preceding angle path base cell insofar as the preceding angle path base cell exists, and are respectively connected with logical "0" in case a preceding angle path base cell does not exist.
  • the output line for the carry bit comes from the next less-significant base cell and is only looped through the angle path base cell.
  • the output bit line for the sum bit is connected to the output of the first register that, in turn, has its input side connected to the sum output of the full adder; and the output line for the carry bit is connected to the output of the second register for the forwarding of the next more-significant angle path base cell, the input side of the second register, in turn, being connected to the carry output of the full adder.
  • each vector iteration stage is composed of a plurality of identical vector path base cells.
  • One vector path base cell has its input side connected to the output lines for carry and sum bits of the immediately-preceding vector path base cell insofar as the immediately-preceding angle path base cell exists and, otherwise, either the input lines for the carry bits or the input lines for the sum bits at the input side of the vector path base cell are connected to the processor input lines for inputting a starting vector.
  • the structure for realizing a shift operation in a vector iteration stage is structured such that a respective angle path base cell has its input side connected to the output lines for carry and sum bits of an i-times more-significant, immediate-preceding vector path base cell and when such a vector path base cell does not exist, has its input side connected to the most-significant, immediately-preceding vector path base cell.
  • a vector path base cell is composed of two multiplexers, of four full adders and of a decoupling device for decoupling the vector iteration stages, the decoupling device being in the form of ten transmission gates.
  • An input line for the sum bit of a first vector component is connected to a first input of a first full adder and an input line for the sum bit of a second vector component is connected to a first input of a second full adder.
  • An input line for the carry bit of a first vector component is connected to a first input of a third full adder via a first transmission gate clocked by a first clock signal and an input line for the carry bit of a second vector component is connected to a first input of a fourth full adder via a fourth transmission gate that is likewise clocked by the first clock signal.
  • An input line for the i-fold more-significant carry bit of the second vector component is connected via the first multiplexer to a second input of the first full adder, the i-fold more-significant sum bit of the second vector component is likewise connected via the first multiplexer to the third input of the first full adder, the i-fold more-significant carry bit of the first vector component is connected via the second multiplexer to a second input of the second full adder and the i-fold more-significant sum bit of the first vector component is connected to the third input of the second full adder, likewise via the second multiplexer, whereby the multiplexers invert or do not invert the carry and sum bits dependent on the sign signals of the immediately-preceding vector iteration stage.
  • the sum output of the first full adder is connected via the second transmission gate to a second input of the third full adder and the sum output of the second full adder is connected via a fifth transmission gate to a second input of the fourth full adder, whereby both of the second and fifth transmission gates are clocked by the first clock signal.
  • An output line for a first-stage carrier bit of a vector component is connected via a third transmission gate to the carry output of the first full adder and an output line for a first-stage carry bit of the second vector component is connected via a sixth transmission gate to the carry output of the second full adder, whereby both of the third and sixth transmission gates are clocked by the first clock signal.
  • An input line for a first-stage carry bit of a vector component from the next less-significant vector path base cell is connected to a third input of the third full adder and an input line for a first-stage carry bit of a vector component for the next less-significant vector path base cell is connected to the third input of the third full adder and an input line for a first-stage carrier bit from the second vector component of the next less-significant vector base path cell is connected to the third input of the fourth full adder.
  • the input lines for the second-stage carry bits of the vector components from the next less-significant vector path base cell are looped through onto the output lines for the immediate-following vector iteration stage of the vector path base cell.
  • the sum output of the third full adder is connected via a seventh transmission gate to the output line for the sum bit of a first component of the immediately-following iteration stage and the sum output of the fourth full adder is connected via a ninth transmission gate to the output line for the sum bit of the second component of the immediately-following vector iteration stage, whereby the seventh and ninth transmission gates are clocked by a second clock signal.
  • An output line for a second-stage carry bit of a first vector component is connected via an eighth transmission gate to the carry output of a third full adder and an output line for a second-stage carry bit of the second vector component is connected via a tenth transmission gate and to the carry output of the first full adder, whereby the eighth and tenth transmission gates are clocked by the second clock signal.
  • the processor is particularly characterized in that at least one vector path iteration stage is redundantly arranged, for example, in order with a given accuracy of the final result vector to allow greater ambiguity regions in the sign detection in the individual angle iteration stages and in order to simultaneously enable an identical correction factor for all combinations of processor input signals.
  • the processor is particularly characterized in that the number of redundantly-arranged vector and angle iteration stages is of such a magnitude that, at most, the foremost-significant carry and sum bits and a sign output signal of the respective immediately-preceding angle iteration stage are required for sign detection in order to enable an optimized sign detector which is identical for all angle iteration stages and which is more simply constructed than a combination of full adders.
  • the processor as featured above is particularly characterized in that an identical, optimized sign detector for all angle iteration stages forms a sign output signal from the foremost-significant carry and sum bits of an immediately-preceding angle iteration stage and from the non-inverted sign output signal of an immediately-preceding angle iteration stage in that the sign output signal of the immediately-preceding angle iteration stage is connected to one of two inputs of an equivalence element and the output of the equivalence element is connected to one of two inputs of a first EXOR (EXCLUSIVE-OR) gate; the most significant carry and sum bits are operated with a second EXOR gate and the output of the second EXOR gate is connected to the second of the two inputs of the equivalence elements; the second most-significant carry and sum bits are operated with a NOR gate, whereby the output of the NOR gate is connected to one of the two OR inputs of a first OR-NAND gate and these carry and sum bits are
  • the output of the first NAND gate is connected to one of the two inputs of a second NAND gate and the output of the second NAND gate is connected to a direct NAND input of the first OR-NAND gate.
  • the respective third most-significant carry and sum bits are operated with the input OR gate of a second OR-NAND gate, whereby the output of the second OR-NAND gate is connected to one of the two OR inputs of a third OR-NAND gate and the third most-significant carry and sum bits are operated by a third NAND gate.
  • the output of the third NAND gate is connected to the respective direct NAND input of the second and third OR-NAND gates.
  • the fourth most-significant carry and sum bits are operated with a fourth NAND gate, the output of the fourth NAND gate being connected to the second OR input of the third OR-NAND gate, and the output of the third OR-NAND gate being connected to the respective second OR input of the first OR-NAND gate and to the second input of the second NAND gate.
  • the output of the first OR-NAND gate is connected to the second input of the first EXOR gate and the output of the first EXOR gate supplies the sign output signal of the optimized sign detector.
  • FIG. 1 is a schematic illustration of the structure for a CORDIC algorithm for vector rotations
  • FIG. 2 is a schematic representation of the ambiguity region in the sign estimation of carry-save numbers
  • FIG. 3 is a block diagram directed to an exemplary embodiment of a CORDIC processor for vector rotations constructed in accordance with the present invention and comprising a vector path and an angle path;
  • FIG. 4 is a schematic representation of an excerpted portion of the angle path of a CORDIC processor constructed in accordance with the present invention and showing angle path base cells and detectors;
  • FIG. 5 is a schematic representation of an excerpted view of the vector path of FIG. 3 of the CORDIC processor constructed in accordance with the present invention showing vector path base cells;
  • FIG. 6 is a schematic representation of an angle path base cell which may be employed in practicing the present invention.
  • FIG. 7 is a schematic illustration of a vector path base cell which may be employed in practicing the present invention.
  • FIG. 8 is a schematic logic circuit diagram of an optimized sign detector which may be employed in practicing the present invention.
  • the idea of the CORDIC algorithm is to execute the rotation of a vector P o (X o , Y o ) by the angle Z o , not in one step, but to approach the rotational angle on the basis of a sum of permanently-prescribed sub-angles alpha i .
  • the sub-angles alpha i With a definition of the sub-angles alpha i as
  • FIG. 1 illustrates the principle of the CORDIC algorithm for vector rotations.
  • the rotation of a vector is therefore realized by a plurality of identical stages that are composed only of adder/subtractor circuits, devices for realizing shift operations and a sign detection of Z i (sign (Z i )).
  • FIG. 2 illustrates the ambiguity region U arising in the sign estimation, showing this ambiguity in a diagram that illustrates the sign (sign result) of the result dependent on the result value RESULT on the word width m and on the number of most significant bits r.
  • FIG. 3 is a block diagram directed to an exemplary embodiment of a CORDIC processor constructed in accordance with the present invention for vector rotations.
  • This processor is composed of a vector path VP, an angle path WP, a multiplication circuit MULK, an adder circuit VMA (vector merging adder) and a clock generator CG.
  • the input quantities of the angle path are the components X o , Y o of the start vector and the input quantity of the angle path WP is the rotational angle Z o .
  • the vector path VP is composed of angle iteration stages IXY 0 . . . IXY 11b; the same holds true of the angle path WP that is of the angle iteration stages IZ 0 . . .
  • each of the angle iteration stages supplies sign signals sign (Z i '), sign (Z i ') to the appertaining iteration stage IXY.
  • sign (Z i ') sign (Z i ')
  • sign (Z i ') sign (Z i ')
  • FIG. 3 illustrates a clock generator CG that generates four clock signals CK 4 from a single clock signal CK.
  • an input word width of, respectively, eleven bits must be provided for the components X o , Y o of the start vector.
  • An internal word width of 17 bits results from the 11 bits of the input word width, one bit for the extreme case that both vector components are added, one bit for a magnification factor 1/K that results on the basis of the iteration and four bits in order to avoid rounding errors.
  • the internal word width must be additionally increased by four places following the decimal point.
  • the output word width for X N ' and Y N ' given the aforementioned accuracy requirement can be respectively reduced to a minimum of 11+1 bits for adding up the vector components, which is equal to 12 bits.
  • the proposals that have been presented proceed from the opinion that errors in the sign recognition should not be made in any stage in order to guarantee the convergency of the CORDIC algorithm.
  • the convergency condition can be more generally formulated as: in case of an incorrect sign decision in the angle iteration stage IZ i, all further angle iteration stages IZ i+1 through IZ n must satisfy the elimination of the error and keep the output Z N within the framework of a given accuracy g.
  • the multiplier circuit MULK can be realized on a hard-wired basis since the multiplication always occurs with a constant correction factor. Due to the redundant stages, a modified correction factor K' results. ##EQU3##
  • the representation in K' in the CSD code enables the realization of the multiplication with only three shift/addition operations corresponding to the significances differing from zero.
  • the adder circuit VMA is executed either as a carry-look ahead adder (CLA) or as a carry-ripple adder (CRA), whereby the carry-look ahead adder CLA has the advantage of a higher processing speed and the carry-ripple adder CRA can be more easily modified in terms of word width.
  • CLA carry-look ahead adder
  • CRA carry-ripple adder
  • the CORDIC processor realized in 1.5 ⁇ m CMOS technology can be operated with clock frequencies up to, typically, 60 MHz.
  • FIG. 4 illustrates an excerpted view of an angle path of the structure of FIG. 3 composed of angle path base cells BCZ and detectors DET, shown in the region of the doubled first angle iteration stages IZ 1a and IZ 1b as well as of the following, regular angled iteration stage IZ 2.
  • an input word Z 1b ' that is likewise composed of carry and sum words is formed in the cells BCZ 1a, 0, . . .
  • BCZ 1a, 12 according to the CORDIC calculating rule, being formed therein from the input words Z 1 a' composed of carry and sum words and the angle step W 1 a that is either inverted or noninverted by the sign signals sign (Z 1 a'), sign (Z 1 a').
  • sign signals sign (Z 1 b'), sign (Z 1 b') for the following angle iteration stage IZ 1b are formed from the four most-significant bits in a detector DET 1a.
  • the angle iteration stages IZ 1a and IZ 1b are identical. Due to the connection to the next iteration stage, a left shift by one bit occurs at the output of the redundant stages, for example, IZ 1b, as well as at the regular stage, this corresponding to a scaling with the factor 2.
  • the wiring of the base cells results from the evaluation of the following equation.
  • FIG. 5 illustrates an excerpted view of the vector path of FIG. 3 in accordance with the invention comprising vector base cells BCXY i ,k in the region of the regular vector iteration stage IXY 2, of the basic stage IXY 3a and of the redundant stage IXY 3b pertaining to the basic stage IXY 3a.
  • the inputs XY 2 ,0 . . . XY 2 ,16 and the outputs XY 4 ,0 . . . XY 4 ,16 are formed of the lines for the carry and sum bits of the vector components X, Y.
  • a respective vector path base cell (BCXY i, k) has its input side connected to the output lines for carry and sum bits (XC i , k, YC i , k, XS i , k, YS i , k) of the immediately-preceding vector path base cell (BCXY i-1, k), insofar as the immediately-preceding vector path base cell exists and, otherwise, either the input lines for carry bits or the input lines for sum bits at the input side given the vector path base cells (BCXY, 0, k) are connected to the processor input lines for inputting a starting vector (X o , Y o ).
  • the structure for realizing a shift operation comprises a respective vector path base cell BCXY i , k with its input side connected to the output lines for the carry and sum bits XC i ,k+1, YC i ,k+i, XS i , k+i, YS i , k+i of an ifold more-significant, immediately-preceding vector path base cell (BXY i-1 , k+i) and, when this vector path base cell does not exist, has its input side connected to the most-significant, immediately-preceding vector path base cell BXY i-1 ,116 (msb).
  • the input word X i must be added to or subtracted from the input word Y i shifted left i times in each vector iteration stage and vice-versa.
  • both data paths for X and Y are bit-by-bit interlaced with one another.
  • FIG. 6 illustrates an angle path base cell BCZ of the present invention.
  • the inputs of the angle path base cell correspond to the inputs of the full adder VA, whereby the one input of the full adder VA is connected to the input lines for an angle bit W i ,k and two other inputs of the full adder VA are connected to the output lines for the carry and sum bits ZC i , k, ZS i , k of the immediately-preceding angle path base cell BCZ i-1 , k insofar as this preceding angle path base cell exists.
  • the inputs of the full adder VA are respectively occupied with a logical "zero" (GND).
  • C 2 MOS registers R1 and R2 are provided, these being clocked by the clock signals CKM and CKS.
  • the clock generator CG mentioned in connection with FIG. 3 generates output signals CK 4, whereby these correspond to the clock signals CKM and CKS and to the clock signals respectively inverted relative thereto.
  • the chronologically-offset clock signals CKM and CKS allow a data transfer according to the master-slave principle.
  • the output line for the carry bit ZC i+1 ,k comes from the next less-significant base cell BCZ i, k-1 and is only looped through the angle path base cell.
  • the output bit line for the sum bit ZS i+1 , k is connected to the output of the first register R1 that, in turn, has its input side connected to the sum output of the full adder VA.
  • the output line for the carry bit ZC i+1 , k+1 is connected to the output of the second register R2 for forwarding to the next most-significant angle path base cell BCZ i, k1, this second register R2 having its input side, in turn, connected to the carry output of the full adder VA.
  • FIG. 7 illustrates a vector path base cell of the present invention that is composed of two multiplexers MUX 1 and MUX 2, four full adders VA1 . . . VA4 and 10 transmission gates 1-10 for decoupling the vector iteration stages.
  • An input line for the sum bit SX i , k of a first vector component X is connected to a first input of the first full adder VA1
  • an input line for the sum bit YS i , k of a second vector component Y is connected to a first input of the second full adder VA2.
  • An input line of the carry bit XC i , k of the first vector component X is connected to a first input of the third full adder VA3 via the first transmission gate 1 clocked by a first clock signal CKM, and an input line for the carry bit YC i , k of the second vector component Y is connected to a first input of the fourth full adder VA4 via the fourth transmission gate 4 that is likewise clocked by the first clock signal CKM.
  • An input line for the i-times more-significant carry bit YC i , k+i of the second vector component Y of the second vector component Y is connected via the first multiplexer MUX1 to a second input of the first full adder VA1, and the i-times more significant sum bit YS i , k+i of the second vector component Y is connected to the third input of the first full adder VA1, likewise via the first multiplexer MUX1.
  • the i-times more-significant carry bit XC i , k+1 of the first vector component X is connected via the second multiplexer MUX2 to a second input of the second full adder VA2 and the i-times more-significant sum bit XS i , k+i of the first vector component is likewise connected to the third input of the second full adder VA2, likewise via the second multiplexer MUX2.
  • the multiplexers MUX1 and MUX2 connect through the carry and sum bits of the immediately-preceding vector iteration stage to the full adder VA1 or, respectively, VA2 either inverter or noninverted dependent on the sign signals sign (Z i '), sign (Z i ').
  • the sum output of the first full adder VA1 is connected via the second transmission gate 2 to a second input of the third full adder VA3 and the sum input of the second full adder VA2 is connected via the fifth transmission gate 5 to a second input of the fourth full adder VA4, whereby both transmission gates 2 and 5 are clocked by the first clock signal CKM.
  • An output line for a first-stage carry bit XCI i+1 , k+1 of a first vector component is connected via the third transmission gate 3 to the carry output of the first full adder VA1, and an output line for a first-stage carry bit YCI i+1 , k+1 of the second vector component is connected via the sixth transmission gate 6 to the carry output of the second full adder VA2, whereby both transmission gates are clocked by the first clock signal CKM.
  • An input line for a first-stage carry bit XCI i+1 , k of a first of the first vector component X from the next less-significant vector path base cell BCXY i, k-1 is connected to the third input of the third full adder VA3, and an input line for a first-stage carry bit YCI i+1 , k from the second vector component Y of the next less-significant vector path base cell BCXY i, k-1 is connected to the third input of the fourth full adder VA4.
  • the input lines for second-stage carry bits XC i+1 , k, YC i+1 , k of the vector component X and Y from the next less-significant vector path base cell BCXY i, k-1 are looped through onto the output lines for the immediate-following vector interaction stage of the vector path base cell VCXY i, k.
  • the sum output of the third full adder VA3 is connected via the seventh transmission gate 7 to the output line for the sum bit XS i+1 , k of a first component of the immediately-following interaction stage, and the sum output of the fourth fill adder VA4 is connected via the ninth transmission gate 9 to the output line YS i+1 , k for the sum bit of the second component of the immediately-following vector iteration stage, whereby the transmission gates 7 and 9 are clocked by the second clock signal CKS.
  • An output line for a second-stage carry bit XC i+1 , k+1 of a first vector component is connected via the eighth transmission gate 8 to the carry output of the third full adder VA3, and an output line for a second-stage carry bit YC i+1 , k+1 of the second vector component is connected via the tenth transmission gate 10 to the carry output of the fourth full adder VA4, whereby the transmission gates 8 and 10 are clocked by the second clock signal CKS.
  • a simple realization of the logic for sign estimating is comprised in a carry ripple adder (CRA) for the four highest significances of Z i ' whose sign is interpreted as the sign of the corresponding carry-save number. Since the four most-significant bits of the scaled angle steps W 0 ' . . . W 11 ', as set forth initially, all begin with the most-significant bits, 11,00, the four-bit ripple adder can be simplified to the optimized detector DET opt illustrated in FIG. 8. The sign estimating therefore occurs parallel to the addition of Z i ' and W i ' in order to therefore shorten the running time of the stage.
  • CRA carry ripple adder
  • a sign output signal (Z i+1 ') is formed from the foremost significant carry and sum bits ZC i , 9, ZS i , 9 ' . . . ZC i , 12 ', ZS i , 12 ' of an immediately-preceding angle interaction stage IZ i-1 and of the non-inverted sign output signal sign (Z i ') of an immediately-preceding angle iteration stage IZ i-1.
  • the sign output signal sign (Z i ') of the immediately-preceding angle iteration state is connected to one of the two inputs of an equivalence gate EQ and the output of the equivalence gate EQ is, in turn, connected to one of the two inputs of a first EXCLUSIVE-OR (EXOR) gate EX1.
  • EXOR EXOR
  • the most-significant carry and sum bots ZC i , 12, ZS i , 12 are operated with a second EXOR gate circuit EX2 and the output of the EXOR gate EX2 is connected to the second of the two inputs of the equivalence gate EQ.
  • the second most-significance carry and sum bits ZC i , 11 ', ZS i , 11 ' are operated with a NOR gate, whereby the output of the NOR gate is connected to one of two OR inputs of a first OR-NAND gate ONA1, and these carry and sum bits are likewise operated with a first NAND gate NA1, whereby the output of the first NAND gate NA1 is connected to one of the two inputs of a second NAND gate NA2, and the output of the second NAND gate NA2 is connected to a direct NAND input of the first OR-NAND gate ONA1.
  • the respective third most significant carry and sum bits ZC i , 10 ', ZS i , 10 ' are operated with the input OR gate of a second OR-NAND gate ONA2, whereby the output of the second OR-NAND gate ONA2 is connected to one of two OR inputs of a third OR-NAND gate ONA3.
  • the third most-significant carry and sum bits are likewise operated by a third NAND gate NA3, whereby the output of the third NAND gate NA3 is connected to the direct NAND input of the second and third OR-NAND gate ONA2, ONA3.
  • the fourth most-significant carry and sum bits ZC i , 9 ', ZS i , 9 ' are operated with a fourth NAND gate NA4.
  • the output of the fourth NAND gate NA4 is connected to the OR input of the third OR-NAND gate ONA3, and the output of the third OR-NAND gate ONA3 is connected to the second OR input of the first OR-NAND gate ONA1 and to the second input of the second NAND GATE NA2.
  • the output of the first OR-NAND gate ONA1 is connected to the second input of the first EXOR gate EX1, and the output of the first EXOR gate EX1 supplies the sign output signal SIGN (Z i+1 ') of the optimized sign detector DET opt .

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Complex Calculations (AREA)
US07/667,289 1990-04-20 1991-03-11 Coordinate rotation digital computer processor (cordic processor) for vector rotations in carry-save architecture Expired - Lifetime US5317753A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE4012709 1990-04-20
DE4012709 1990-04-20

Publications (1)

Publication Number Publication Date
US5317753A true US5317753A (en) 1994-05-31

Family

ID=6404786

Family Applications (1)

Application Number Title Priority Date Filing Date
US07/667,289 Expired - Lifetime US5317753A (en) 1990-04-20 1991-03-11 Coordinate rotation digital computer processor (cordic processor) for vector rotations in carry-save architecture

Country Status (4)

Country Link
US (1) US5317753A (de)
EP (1) EP0453641B1 (de)
JP (1) JPH04227568A (de)
DE (1) DE59010674D1 (de)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5889822A (en) * 1995-08-30 1999-03-30 U.S. Philips Corporation Signal processor with reduced complexity, and receiver comprising such a signal processor
US6349317B1 (en) * 1999-03-13 2002-02-19 Vitit Kantabutra Efficient radix-4 CORDIC vector rotators and computers of sine and cosine functions
US6385632B1 (en) 1999-06-18 2002-05-07 Advanced Micro Devices, Inc. Fast CORDIC algorithm with sine governed termination
US6434582B1 (en) 1999-06-18 2002-08-13 Advanced Micro Devices, Inc. Cosine algorithm for relatively small angles
DE10164462A1 (de) * 2001-12-20 2003-07-10 Ihp Gmbh CORDIC-Einheit
US6693970B2 (en) * 1999-04-23 2004-02-17 Nokia Corporation QAM modulator
US20060090060A1 (en) * 2002-02-18 2006-04-27 Wolfram Drescher Method and arrangement for bringing together data on parallel data paths
US20070130241A1 (en) * 2005-11-23 2007-06-07 Leo Bredehoft Systems and methods for implementing CORDIC rotations for projectors and related operators
US20090327667A1 (en) * 2008-06-26 2009-12-31 Qualcomm Incorporated System and Method to Perform Fast Rotation Operations
US20110225222A1 (en) * 2010-03-15 2011-09-15 Integrated Device Technology, Inc. Methods and apparatuses for cordic processing
CN108416311A (zh) * 2018-03-14 2018-08-17 电子科技大学 一种基于可编程门阵列和坐标旋转处理的方位角获取方法

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE4126953C2 (de) * 1991-08-14 1995-04-13 Fraunhofer Ges Forschung Schaltungsanordnung zur Durchführung des CORDIC-Algorithmus
DE4313627C1 (de) * 1993-04-27 1994-05-05 Ant Nachrichtentech Verfahren zum Ermitteln der Eigenwerte einer Matrix aus Signalabtastwerten
US6253526B1 (en) 1998-11-13 2001-07-03 E.I. Du Pont De Nemours And Company Installation method for carpet underlays
US20060200510A1 (en) * 2005-01-31 2006-09-07 Wang Cindy C Precision cordic processor
RU2475830C2 (ru) * 2010-08-13 2013-02-20 Виктор Николаевич Бабенко Устройство вращения вектора

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4231102A (en) * 1978-12-21 1980-10-28 Raytheon Company Cordic FFT processor
US4337519A (en) * 1979-02-01 1982-06-29 Tetsunori Nishimoto Multiple/divide unit
US4757444A (en) * 1983-12-26 1988-07-12 Hitachi, Ltd. Vector processor capable of performing iterative processing
US4791590A (en) * 1985-11-19 1988-12-13 Cornell Research Foundation, Inc. High performance signal processor
US4896287A (en) * 1988-05-31 1990-01-23 General Electric Company Cordic complex multiplier
US4945505A (en) * 1988-10-17 1990-07-31 Raytheon Company Cordic apparatus and method for approximating the magnitude and phase of a complex number
US5019968A (en) * 1988-03-29 1991-05-28 Yulan Wang Three-dimensional vector processor

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4231102A (en) * 1978-12-21 1980-10-28 Raytheon Company Cordic FFT processor
US4337519A (en) * 1979-02-01 1982-06-29 Tetsunori Nishimoto Multiple/divide unit
US4757444A (en) * 1983-12-26 1988-07-12 Hitachi, Ltd. Vector processor capable of performing iterative processing
US4791590A (en) * 1985-11-19 1988-12-13 Cornell Research Foundation, Inc. High performance signal processor
US5019968A (en) * 1988-03-29 1991-05-28 Yulan Wang Three-dimensional vector processor
US4896287A (en) * 1988-05-31 1990-01-23 General Electric Company Cordic complex multiplier
US4945505A (en) * 1988-10-17 1990-07-31 Raytheon Company Cordic apparatus and method for approximating the magnitude and phase of a complex number

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Despain, Alvin M., "Fourier Transform Computers Using CORDIC Iterations", IEEE Transactions on Computers, vol. C-23, No. 10, Oct. 1974, pp. 993-1001.
Despain, Alvin M., Fourier Transform Computers Using CORDIC Iterations , IEEE Transactions on Computers, vol. C 23, No. 10, Oct. 1974, pp. 993 1001. *
Hahn et al., "Cordic-Prozessover fur die digitale Signzlverarbeitung", ME, vol. 3, No. 1, 1989, pp. 22-27.
Hahn et al., Cordic Prozessover f r die digitale Signzlverarbeitung , ME, vol. 3, No. 1, 1989, pp. 22 27. *
Haviland et al. "Acordic Arithmetic Processor Chip", IEEE Transactions on Computers, vol. C-29, No. 2, Feb. 1980, pp. 68-79.
Haviland et al. Acordic Arithmetic Processor Chip , IEEE Transactions on Computers, vol. C 29, No. 2, Feb. 1980, pp. 68 79. *

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5889822A (en) * 1995-08-30 1999-03-30 U.S. Philips Corporation Signal processor with reduced complexity, and receiver comprising such a signal processor
US6349317B1 (en) * 1999-03-13 2002-02-19 Vitit Kantabutra Efficient radix-4 CORDIC vector rotators and computers of sine and cosine functions
US6693970B2 (en) * 1999-04-23 2004-02-17 Nokia Corporation QAM modulator
US6385632B1 (en) 1999-06-18 2002-05-07 Advanced Micro Devices, Inc. Fast CORDIC algorithm with sine governed termination
US6434582B1 (en) 1999-06-18 2002-08-13 Advanced Micro Devices, Inc. Cosine algorithm for relatively small angles
DE10164462B4 (de) * 2001-12-20 2007-04-26 IHP GmbH - Innovations for High Performance Microelectronics/ Institut für innovative Mikroelektronik GmbH CORDIC-Einheit
US20060059215A1 (en) * 2001-12-20 2006-03-16 Koushik Maharatna Cordic unit
DE10164462A1 (de) * 2001-12-20 2003-07-10 Ihp Gmbh CORDIC-Einheit
US7606852B2 (en) 2001-12-20 2009-10-20 IHP-GmbH-Innovations for High Performance Microelectronics/Institut fur Innovative Mikroelectronik CORDIC unit
US20060090060A1 (en) * 2002-02-18 2006-04-27 Wolfram Drescher Method and arrangement for bringing together data on parallel data paths
US7779229B2 (en) * 2002-02-18 2010-08-17 Nxp B.V. Method and arrangement for bringing together data on parallel data paths
US20070130241A1 (en) * 2005-11-23 2007-06-07 Leo Bredehoft Systems and methods for implementing CORDIC rotations for projectors and related operators
US7818357B2 (en) * 2005-11-23 2010-10-19 Rambus Inc. Systems and methods for implementing CORDIC rotations for projectors and related operators
US20090327667A1 (en) * 2008-06-26 2009-12-31 Qualcomm Incorporated System and Method to Perform Fast Rotation Operations
US8243100B2 (en) 2008-06-26 2012-08-14 Qualcomm Incorporated System and method to perform fast rotation operations
US20110225222A1 (en) * 2010-03-15 2011-09-15 Integrated Device Technology, Inc. Methods and apparatuses for cordic processing
US8572151B2 (en) 2010-03-15 2013-10-29 Integrated Device Technology, Inc. Methods and apparatuses for cordic processing
CN108416311A (zh) * 2018-03-14 2018-08-17 电子科技大学 一种基于可编程门阵列和坐标旋转处理的方位角获取方法

Also Published As

Publication number Publication date
EP0453641B1 (de) 1997-03-12
JPH04227568A (ja) 1992-08-17
EP0453641A3 (de) 1995-04-19
DE59010674D1 (de) 1997-04-17
EP0453641A2 (de) 1991-10-30

Similar Documents

Publication Publication Date Title
US5317753A (en) Coordinate rotation digital computer processor (cordic processor) for vector rotations in carry-save architecture
US6230179B1 (en) Finite field multiplier with intrinsic modular reduction
US5798955A (en) High-speed division and square root calculation unit
US5023827A (en) Radix-16 divider using overlapped quotient bit selection and concurrent quotient rounding and correction
US4683548A (en) Binary MOS ripple-carry parallel adder/subtracter and adder/subtracter stage suitable therefor
US5010511A (en) Digit-serial linear combining apparatus useful in dividers
US6370667B1 (en) CRC operating calculating method and CRC operational calculation circuit
JP3003467B2 (ja) 演算装置
US5574677A (en) Adaptive non-restoring integer divide apparatus with integrated overflow detect
US4866655A (en) Arithmetic processor and divider using redundant signed digit
US5084834A (en) Digit-serial linear combining apparatus
RU2348965C1 (ru) Вычислительное устройство
US6484193B1 (en) Fully pipelined parallel multiplier with a fast clock cycle
US3489888A (en) Floating point look-ahead binary multiplication system utilizing two's complement notation for representing negative numbers
US8417761B2 (en) Direct decimal number tripling in binary coded adders
He et al. Stochastic circuit design based on exact synthesis
US5978826A (en) Adder with even/odd 1-bit adder cells
US5719798A (en) Programmable modulo k counter
KR100241071B1 (ko) 합과 합+1을 병렬로 생성하는 가산기
RU2799035C1 (ru) Конвейерный сумматор по модулю
Surarerks Digit set conversion by on-line finite automata
US5381380A (en) Divide circuit having high-speed operating capability
Nagendra et al. Unifying carry-sum and signed-digital number representations for low power
JP2705162B2 (ja) 演算処理装置
RU2231823C2 (ru) Устройство для контроля позиционных сумматоров по модулю

Legal Events

Date Code Title Description
AS Assignment

Owner name: SIEMENS AKTIENGESELLSCHAFT, A GERMAN CORPORATION

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNORS:KUENEMUND, RONALD;NOLL, TOBIAS;REEL/FRAME:005636/0021

Effective date: 19910222

STCF Information on status: patent grant

Free format text: PATENTED CASE

FEPP Fee payment procedure

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 4

FPAY Fee payment

Year of fee payment: 8

FEPP Fee payment procedure

Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 12

AS Assignment

Owner name: INFINEON TECHNOLOGIES AG, GERMANY

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SIEMENS AKTIENGESELLSCHAFT;REEL/FRAME:026358/0703

Effective date: 19990331